Fast clustering based on state learning machine
Kou Yu, Qingxiang Wu, Xue Li, Sanliang Hong · 2016
Clustering has widely been applied in various domains such as artificial intelligence and big data processing. The main task of clustering is to partition a data set into different classes or clusters according to a certain standard (such as distance criterion), which makes the similarity of the data objects in the same cluster as large as possible, while the difference between different clusters is as large as possible. Similarity measurement is the key to ensure that the similarity between the two elements in the same cluster is larger than the similarity between the two elements from different clusters, and also is larger than the similarity between the two clusters. In order to efficiently measure similarity and fast clustering, the State Learning Machine (SLM) is proposed in this paper. In the SLM two elements in the same cluster are regarded as a friend pair, therefore the clustering process becomes a novel approach of gathering friends. In the condition of large cluster differences, the SLM is not only able to cluster data fast and correctly with very low complexity, but also effectively detect the isolated points and noise points. It is not sensitive to the input order of noises and outlier data, furthermore, it can discover clusters with arbitrary shapes and can be applied to cluster big data efficiently.